Nvidia's been pretty terrible for open source / free software. No need to quote Linus Torvalds here. They want to control what runs on their hardware. They want to you write code against their proprietary drivers and APIs, not directly against the hardware (which these days of course also contains plenty of software, but still).
Don't expect things to go differently this time around. Nvidia wants control over the software stack. Acquiring HF fits in perfectly. The play is long term.
Yes, there is no question NVIDIA wants to lock you into CUDA and their hardware. But also, they’ve consistently demonstrated the most openness when it comes to model training, datasets, and research; even before the LLM era (e.g. StyleGAN).
There’s also modelscope.cn (china’s huggingface) which is worth checking out. I would not be surprised if one day, we have to use China VPNs to download open weight models.
"The proprietary shovel seller has some excellent tutorials on how to dig gold. Nobody else has such good step by step guides. Therefore them buying a shovel-agnostic tutorials and techniques method (that has a lot of info on using other shovels effectively) is justified."
Evidently we should, because Linus has been more positive about Nvidia in the last 2 years [0]. I've been using the open driver for years now, for both gaming and CUDA.
He definitely sounds more pragmatic than before, which is, in a sense, more positive.
> This is actually one of the benefits brought by AI; it has made Nvidia a good participant in the Linux kernel space. [...] Now, when Linux is so important for AI clouds, Nvidia suddenly cares very much about Linux.
Nvidia releases some of the most open open weights models, Nemotron 3, which have the full training code open, and most but not all of the training datasets.
Nvidia is a big company. They are good about some things and bad about others.
I think they really do like open weights because they make some of the best hardware for training, and the more open weights models there are, the more people are training and fine-tuning them, mostly on Nvidia hardware.
I feel like Nvidia is one of the better choices for buying Huggingface. Not perfect, but definitely far from the worst.
I think this is part of an open-source play. I'm not arguing your other points, I think they're true.
They're trying to mix up the competitive landscape(that doesn't impact their bottom line, and I don't think opensource is eating their lunch), so I don't think this is fake, at least that's my initial take.
> as a force to have open weight models run better on Nvidia against the rest.
this is the crux - if nvidia makes it so that open weights end up running better on nvidia hardware than competitor's, then it's going to prevent hardware innovation and competitiveness in the entire sector.
It's like as tho General Motors buys out oil refinery to make gas for all, but the gas somehow runs smoother in GM cars.
It's a clever move if that's what they're doing. They're restricted in China, and are likely to face stiff competition from Chinese chipmakers in the coming years. Acquiring the largest repository of trainable models and ensuring they run better on Nvidia hardware is probably one of the few moves they have for keeping ahead of the competition. I mean, it would be terrible for the consumer, but it does make me think that NVDA is a decent investment.
We can stop with these weak excuses since AMD and Intel have done more for open source than Nvidia has, including their GPU drivers for Linux.
Nvidia on the other hand has not and the best they have done is a bunch of closed-source blobs which they do more closed source releases than the rest.
Mojo is open source and targets all GPU architectures for their compiler regardless of the vendor and nvcc targets their own (and both that and CUDA are closed source).
So this is directly an apples to apples comparison.
"That's not how citations work." (Dude on the interwebz, 2026)
But more seriously, this is my first time seeing that as well, and I'm not sure I like it. Citing an LLM is a little like citing Wikipedia to me, you cite the primary source the LLM is quoting directly, not the secondary source.
It’s the formatting and formality that bother me. Say “I looked it up with Gemini,” don’t give me some weak attempt at “proper citation” to legitimize the bare minimum effort you put in.
They’re buying a brand, some employees, and some momentum — not any of their software. HF probably does have propriety goodies to make it all run efficiently, but certainly not a billion dollars worth, much less 13!
We are so used to these numbers being thrown around in the AI era that something one needs a reminder that this is 13 billion, not million. Insane exit by HF.
Was looking for this comment, the tech world has gotten completely insane with ”valuations”. I would love to hear why it was 13 and not 5. It would still be completely insane at 5 billion, but someone though they should add another 8…
Potentially horrible for monopoly reasons, but if other big acquisitions in the AI era teach us anything, developers are about to get a whole lot of free and discounted trial credits.
That’s at least a plus. I will happily burn through as much VC money as they will give me to tinker with my projects.
Now that you mention sleeping. I imagine one could do cool sleeping startups that clone something but exist only on paper. The sleeper startup is only launched the moment some ham fisted mega corp aquires the original and all the customers are looking for a way to abandon ship.
I think failing to take advantage of the charisma of both Lina Khan and Janet Yellen, and not making them public-facing economics gurus like Alan Greenspan or Larry Summers historically were, was a huge tactical mistake by Biden.
Yellen was objectively the dumbest Treasury Secretary ever. When everyone and their dog realized inflation was getting out of control, she thought it was transitory. She literally gets an F- as an economist.
And some former Roomba employees may have thoughts about Kahn.
Kahn was right about Roomba. Amazon was going to torpedo all the superior competition on the Amazon platform where almost all robot vacuums are sold in the US, plus as the e-commerce monopoly, Amazon was going to instantly have access to an army of cameras inside people’s homes to analyze personal information related to purchase preferences.
IMO allowing Roomba to hit chapter 11 was still a better option for the consumer than handing them to Amazon. They’re still in business as an independent competitor on the market, consolidation was successfully avoided.
Yellen, I don’t have much of an opinion on, but she absolutely wasn’t alone in that opinion (the nuance of that opinion being inflated by this hyperbole), and the treasury is a lot less involved than the federal reserve in doing anything about inflation, anyway.
Roomba was not competitive in the market, no amount of Amazon market manipulation would have changed that. Best case scenario, Amazon would have invested in making it competitive. Worst case, their engineers would have been absorbed into Amazon's warehouse robotics projects.
Is either of those worse than what actually happened: the company is now a zombie brand for a Chinese company?
Of course Amazon would have changed that. You would go on Amazon and search for "robot vacuum" and Amazon would put iRobot at the top of the results. Review manipulation on the platform would be trivial.
Amazon could email/push notification/text customers asking for reviews of iRobot vacuums but then not do the same for competing vacuums, skewing their reviews higher (asking for reviews boosts ratings by gathering opinions from happy customers who usually don't bother writing a review). Competing brands potentially don't even have your contact information to ask for a review.
Go on Amazon right now and search for "usb cable." There's a giant banner at the top that recommends the Amazon Basics brand, three across horizontally, which on my desktop monitor takes up nearly 50% of the screen real estate. Then below it are the Anker cables that you're more likely to be looking for.
They would have almost certainly been manipulating pricing on them as well. For example, they could take the strategy of lowering the price of the vacuums to break even or sell as a loss leader, but use them as a data-gathering robot in your house to help Amazon sell more of everything else. They could make their Alexa smart home platform preferential to iRobot or lock out competing models.
Robot vacuums are a consumer goods category that is heavily skewed toward Amazon.com as the place of purchase compared to other retailers.
I think "zombie brand owned by a Chinese company" is actually preferable, yes. They still operate and sell vacuums competing with the other robot vacuums on the market, and they aren't in service as household data collection bots for a monopoly e-commerce platform.
I don't think the ownership of the company being foreign or domestic is very relevant to the FTC's goal of preserving positive trade conditions. Would we think the iRobot situation was a bad outcome if iRobot was purchased by a foreign company we view more positively like Miele? We only think of it negatively due to anti-Chinese bias. iRobot being Chinese-owned is almost certainly the best possible outcome for preserving the amount of competition in the market.
This is a tangent to my first comment parent to this one, but also more related to the actual article at hand: whether it's Amazon buying Roomba or Nvidia buying Hugging Face, I have a somewhat radical (or is it?) opinion that large companies like Nvidia, Amazon, Apple, Microsoft, Coca-Cola, etc, should not actually be allowed to acquire companies. At all. under any circumstances, even if competition is healthy.
These companies are generally large enough that they do not need the competitive aid of buying an existing company and starting with that sort of structural advantage.
E.g., did Nvidia not have enough money in their bank to start a company to compete with Hugging Face? This is a company that reportedly has ~200-300 employees with investment rounds totaling $400 million. Nvidia made $31.9 billion in net income last quarter.
I look at a company like Xiaomi which just developed its automotive division in-house without resorting to buying car companies. I think our traditional business and finance mindset has an overreliance on acquisitions.
Yes, that's a very good company to be the owner of iRobot, because they are just a robot factory and not a near-monopoly e-commerce retailer.
They are essentially on equal footing with other robotic vacuum manufacturers. iRobot didn't go out of business or get absorbed into a larger company lowering competition in the marketplace.
It's also not the FTC's job to ensure that companies, especially ones with zero/trivial national security or domestic labor force value, remain under domestic ownership. Amazon itself is not really a "domestic" company, it's publicly traded. Anyone from any non-sanctioned country can buy shares of Amazon.
I spent a little time working there around 2017. Whatever you're implying my former colleagues' thoughts on the subject are (or mine, for that matter), you're probably wrong.
True, but better nvidia than any other big tech contender. Nvidia has a very strong business incentive to make huggingface thrive, whereas pretty much everyone else has the opposite incentive.
The Government plans to regulate open source ai. Nvidea is dependent on the Government to back loans, provide property, and make laws in their favor. Now they have ownership of what the Government wants to control. A perfect wedge they can use to get more from the Government.
Why would China want to crash the AI industry? Their goal is not to damage the United States, necessarily. Their goal is to have the best domestic AI on the planet. They intend to do that through the way that they dominate every other industry: good enough quality at a much greater scale.
I think this is good.. They have incentives to keep things free to keep people using their GPUs. I think that's one of the least enshittifying outcomes possible
The $13b doesn't go to hugging face's bank acct to pay for anything like S3 egress fees... It go to hugging face's owners' Bank accounts for them to do anything else. New cohorts of billionaires and centi millionaires getting minted.
Obviously, NVIDIA is trying to own the AI development chain.
Owning HF -- the discovery and distribution channel -- is one thing, but I think the biggest threat vector is the privileged access to HF platform data, that includes HW survey info and model download pattern. This can be a borderline anti-trust case.
The more interesting is HF turned down a $500M Nvidia investment late last year at
a $7B valuation, after passing on a $235M round in 2023 at $4.5B —
going from "we don't want a dominant investor" to a $13B full
acquisition in under a year is quite the reversal.
I don't think they passed on the 4.5 billion. They still raised it in 2023 with investors like Alphabet etc. They turned Nvidia *earlier this year* for the $7 billion valuation deal. It makes sense if they were angling for an acquisition.
Reversal would be "valued at 7B then valued at nothing", this is more like "nah, we want more" then "nah, we want more" then "yes, that's what we want :)".
I'm sure it'll be fine. As long as you have the latest 90-class card you'll probably continue have nearly full functionality from the HF libraries. (Upgrade to enterprise gear for full functionality)
The fit is theoretically good (nvidia is model agnostic) but I can see then pushing the ecosystem increasingly towards cuda and some custom nvidia software packages for inference etc etc
The unlimited private repos for free tier would have happened anyways. You have to remember that this was around the time people were jumping ship from GitHub exactly because they didn't offer private repos for free. It was something many wanted, and they were actively seeking out other places that would give it to them.
Maybe but they make their real money in data centers. Certainly, they want to have fingers on every trend. Owning something the Government wants to control gives you more access to trade repression of open models for more Government favors and money.
I’m asking more to work out what is the basis of this valuation. Why would Nvidia spend $13B for what seems to be a services that gives things away for free.
I dunno. You can argue over whether they're overpaying, but it's not like Huggingface is Clinkle. They hit $150 million in ARR this year, they have tons of runway, and according to reports, have just started to even burn the money they raised a few years ago.
I get that it's fun to be glib about the stupidity of tech elites and investors in general, but Huggingface have been pretty open about their financials and are, in my opinion as a practitioner in the field, one of the most responsible orgs in our space. They've been a pillar of open source ML for years now and have made a very positive impact on our ecosystem.
Nvidia is getting a real business generating revenue, and the center of the universe for open models. Both seem like pretty valuable attributes, from Nvidia's perspective.
Solid point. I'm sure Nvidia went into this deal expecting completely flat growth and no other benefits to their core business. Sorta like how Meta never increased Instagram's revenue from $0 and is still waiting for it to pay off that billion dollar acquisition price.
Or like GitHub, which was generating something like 200 million in ARR and had never hit profitability when Microsoft bought it for $7.5 billion back in 2018. I'm sure it has come as nothing but a happy surprise to Microsoft that GitHub generated $1 billion in 2023. They had initially penciled it in for 38 years til ROI.
Nvidia has a market cap of $5T USD today, and a decent chunk of that is due to LLM speculation.
Does Nvidia want their stock price to be at risk of being tanked by a download service being in the news? No, they want to make sure the party keeps going and is under their direct supervision, and part of that is making sure Hugging Face isn't bought by a competitor or runs out of money.
Yeah, great point. Nvidia could potentially be overpaying. Not sure how that equates to Huggingface being "a file download mirror with a couple of side features dangling off".
Absolutely none of this makes sense and the thing that amazes me is how long it has continued. Future historians will just laugh at how stupid and obvious the crash was.
While it's probably easier to say this in retrospect, they were eliminating a direct competitor to their core business. Something so advantageous it should have been blocked by regulators.
I can't see this as being as good of a purchase, especially when it's 13x the price of what was seen as an absurdly large amount back then.
They have various paid services that relate to AI—paid inference hosting; paid accounts aimed at AI development with GPU rentals [credits plus paid overages, I believe], more private storage and public storage than free accounts; and many of the community a and some other benefits on individual/team/enterprise tiers; additional paid storage above the base quotas for the paid account tiers; on demand rentals of HF managed containers on GCP and AWS, and some other things.
I guess we may have to start paying to download models.
Or perhaps they will start throttling downloads for free users.
I don't know what they business case is, it might be to shut them down: I suspect good free models on local hardware is a threat to Nvidia's investments in OpenAI/Anthropic.
NVidia has been expending energy helping improve local models and inference platforms for them targeting NVidia GPUs; good free models that users can run locally rewards Nvidia’s investment in product lines for local inference (DGX, RTX PCs, etc), as well as—given their continued dominance in the space—the premium over competitors of their consumer and workstation GPUs.
Maybe they're trying to see how big the market actually is before considering shutting it down, maybe or getting lawyers and politicians to try and outlaw or restrict open models if they see a big enough opportunity.
The open weights AI ecosystem is way too concentrated with HF. That said, it was clear they were going to either get bought or aggressively monetized at some point, Nvidia seems like it could be one of the more aligned acquirers.
I do worry about any sort of crowding out or downplaying non Nvidia-relevant quants, and about changing rules to crack down on models or datasets that for one reason or another “don’t align with their corporate values” - uncensored etc. Someone mentioned Microsoft and GitHub, they appear to me anyway to have been very hands off, I hope it’s the same model.
It's already part of Alibaba, which has its own incentives. So does kaggle with Google. Civitai is neutral but it is mostly nsfw even though they don't show it outright. Torrenting is the Best way
HF is the default platform to find models. Not just ones published by big labs but also a lot of the distilled or fine-tuned versions, etc.
If Nvidia buying HF makes it tough for all the diverse models on HF, then what are some alternatives?
It seems models are the best things to be available on a Torrent platform? Of course HF is much more than just the files but perhaps the metadata can be separate and hosted on multiple community platforms.
I'd honestly prefer that. I'm usually on 5g if not 4g and pulling models is an absolute nightmare when I basically have to use HF and even with a token getting throttled. Although I have a pretty unique use case. Pulling wads of 5gb tensor files over cellular is painful.
I am on the same boat, I live in a small Himalayan village. I have 2x5G based devices and one Wireless bridge (Ubiquiti LiteBeam M5) for a local broadband. Generally I get 50 Mbps, sometimes up to 100 Mbps
This is the world we live in now. 3-4 megacorps owning everything in every major product category - eyewear, cosmetic, consumer goods, media, tech; you name it.
We need small companies worldwide, developing the things huggingface does. The us china and Europe dominate while the rest of the world sits by idly. Torrent and seed ai models. It's not piracy
Same reason why GitHub was acquired for $7.5B at $250 million ARR which is 30x revenues:
VCs could not see any other reason to raise more money and Huggingface was not growing as fast as they thought to justify the valuation or the next fundraise.
So they might as well get Nvidia to save them from the VCs pressurizing them.
That doesn't make much sense to me. Companies don't just buy companies for absurd multiples out of the niceness of their heart.
In the case of GitHub, it was likely for data reasons + wanting to own where developers do work (VScode + Github).
In the case of HuggingFace, honestly not sure as I'm not familiar enough with their business. But I can assure you that Nvidia didn't buy them for 13 billion cause HuggingFace were desperate. When you're desperate, you sell for less not more.
I remember when people realized gaming cards could be used for general compute (GPGPU) on certain types of parallel numeric work. It was a very cool thing for many types of hackers and science and engineering students and NVIDIA went far out of their way to ensure we could never buy it.
The optimal market strategy there (as in a lot of places) wasn't "sell as much as you can". There's often a superior strategy, when (as with HPC) you have minority industry customers who are very rich and have low price sensitivity. It's to raise the price to what those special customers are willing to pay, and to drop everyone else.
What NVIDIA did was to rip out FP64 capability, systematically, from all of their consumer cards. They firewalled off "useful for GPGPU" as a differentiating feature, segmented the market, and astronomically raised the price of what (if you were looking soley at cost-to-manufacture) could have been easily affordable to any ramen student.
(It's a more obscure version of the Intel-made-ECC-memory-disappear story).
The same is true for the simultaneous NVENC video encoding session limit which is artificially locked for consumer NVIDIA GPUs (see https://github.com/keylase/nvidia-patch)
Yes, this is ironic - it was the tinkerers using CUDA on cheap graphics cards that made Nvidia stuff useful in constrained academic environments.
True about Intel and ECC, but AMD now does similar things, even with their consumer CPUs and chipsets.
These three companies now make very sure that consumer products can never canibalize those juicy data center profits - so they make sure to limit what the consumer segment can do.
Vertically integrated monopolies naturally have aligned in incentives for themselves, which allows more capture which turns out bad for customers. Whether it is a monopoly is an open question.
This isn't vertical integration. This is a commoditize your complements play and that does benefit customers. Nvidia benefits from supporting the open model ecosystem, if models are a commodity the gpus become what's scarce.
Nvidia is indeed a terrible company when it comes to open source, etc. however, as someone who has been gaming on desktop hardware since the 6xxx card days, Nvidia hardware has always worked. I switched to Linux full time since the 2XX days (maybe it was the 280 gtx?) and never had a lot of the problems everyone else got on Linux (using only the closed source drivers).
NVIDIA introduced programs like The Way It's Meant to be Played, optimizing major game titles specifically for NVIDIA architecture, which made competitions like AMD cards underperform in critical releases.
But they don't/didn't have the same level of market share as Nvidia, particularly in the PC gaming market. Nvidia had 65%-85% market share, and still worked with game studios to make sure their competitions failed. That's the opposite of good stewardship that's all I'm saying.
They worked with studios to optimize games for their hardware. I don’t think they went around sabotaging other companies hardware. They did what was good for them.
Likewise, their hardware has been rock solid and very performant for me. I bought most of my current setup second hand and they've been going strong for years now.
I think they will be, Nvidia has released open source model that also included the data it was trained on. I think they are the only ones that have done that.
They were already pretty vocal about it, but this makes Nvidia pretty much the de facto face of open weight models.
As battle lines get drawn over duopoly vs. open weight it’ll be interesting to see what Nvidia does. They definitely want a piece of more of the stack especially as Huawei chips become more and more of an alternative to cuda.
It was either they (literally) sell out or enshittify. Choose your poison. Hosting and running LLMs is an expensive business to be in. There wasn't ever going to be a good ending to this story. The original investors need to be made whole.
Not if they can't indefinitely contain ram and SSD procurement. And they can't and the Market already showed with specex that they are unwilling to float these fly by night outrageous 'investments' onto institutional holders.
Institutional holders mass revolted at spacex getting into the basket.
AI costs more than having people do the work. Q2 CFO reaction proved that.
I'm puzzled by the reactions. This is not a good thing. Yes, they're releasing more open source code and models, but they do it BECAUSE OF pressure from the community and other open source projects.
But Nvidia is a terrible open source and consumer company. They gatekeep a lot and oftentimes it's only open source in name. Outside contributions are often slow-walked or rejected if they don't align with business incentives , and leadership is retained 100% in a couple of people from a certain country.
Linus isn't wrong but at the same time NV has shipped more compute than AMD and Intel combined to consumers. The kind of performance I get out of a modest outlay never ceases to amaze me.
NV is a very different beast from all of those. I have yet to regret buying any of their products and I wouldn't touch your list with a stolen 10' pole.
Isn't it super easy for the community to just replace HuggingFace? Isn't it pretty much just a repo/index for open-weight models? Why is it even worth anything?
> Isn't it pretty much just a repo/index for open-weight models?
No, its not just a repo/index (they also have training, inference hosting, and they develop/maintain a bunch of core AI infrastructure software), and even if it was just a repo/index, replacing a bug centralized repo/index that used by an large community isn’t trivial.
Nvidia has already a good free offer with Nim and a lot of compute power available. Plus, they obviously have interest in spreading open models culture.
We keep talking about EU having to be more present in the AI race, but if the lifecycle of European AI companies is to be bought by US ones, what’s the point?
HuggingFace offers abliterated/uncensored models, which would seem to run contrary to the "AI is too powerful for regular people" narrative that the big players are pushing.
My primary concern is that Nvidia will bow to the pressure and restrict abliterated/uncensored models on HF.
I really wish I'd done better bookmarking back when Nvidia was talking about acquiring ARM. There were all kinds of things coming out saying, if we do this, even though we will be taking over ARM, it will transform us from the inside out. We would become a new different company, that cares about something beyond our own self interest, our own chips.
This feels like a similar leap of faith. One that is hard to believe in. Thankfully, I think Nvidia can keep the lights on here & keep this going. I don't think they have to do much, per se. But it felt implausible then to image an Nvidia that gave a shit about anyone else, an Nvidia that actually gave a flying fuck about drivers or upstream Linux or ecosystems that weren't entirely within their own control.
Similarly the upper quartile of succes here feels mostly like benevolent neglect. I think we can hope for Nvidia to just not mess up a good thing, for them to understand that this open model open ai universe hinges upon Hugging Face, and for them to pretty please keep caring about the existential risk of the hyper-ai'ers all building their own properietary models on proprietary hardware and leaving Nvidia behind some day, and HF being the hedge against being left behind.
The story is in an intermediate state right now, but as semiquaver points out, The Information is reporting it as fact. Since their reporting tends to be as reliable as it is hardwalled, we went with their claim in the title.
Please don't post unsubstantive comments to Hacker News.
No doubt you have the nucleus of a substantive comment here, but that's not enough. If you only post the shallowest top stratum of what you're thinking, other people do the same, and then we get "Laws are for poor people" and endless descending repetition. The whole point of this site is to try for something other than that.
I thin HF holds huge power in how they’ve consolidated all the open models and data, but monopoly feels like a stretch here. They are in reality a fairly small startup with lots of alternatives (all of which are much worse at this point granted) and are mostly a loss leading part of the open source ecosystem. I think it would be very hard to make a case against this on monopoly grounds, though I wouldn’t be opposed to someone trying if they thought it was possible.
Nvidia's been pretty terrible for open source / free software. No need to quote Linus Torvalds here. They want to control what runs on their hardware. They want to you write code against their proprietary drivers and APIs, not directly against the hardware (which these days of course also contains plenty of software, but still).
Don't expect things to go differently this time around. Nvidia wants control over the software stack. Acquiring HF fits in perfectly. The play is long term.
NVIDIA is one of the most open labs.
They even share many of their pre-training and even post-training datasets for Nemotron on HuggingFace; for example: https://huggingface.co/datasets/nvidia/Nemotron-Post-Trainin...
Which other lab shares this?
Yes, there is no question NVIDIA wants to lock you into CUDA and their hardware. But also, they’ve consistently demonstrated the most openness when it comes to model training, datasets, and research; even before the LLM era (e.g. StyleGAN).
There’s also modelscope.cn (china’s huggingface) which is worth checking out. I would not be surprised if one day, we have to use China VPNs to download open weight models.
> NVIDIA is one of the most open labs.
Of course they are. They're commoditizing their complement.
I want to own the hardware, not play around in an nvidia fiefdom full of nvidia rules.
nothing stopping you from doing that
Can't buy a decent modern Intel gpu to run AI on because of nividias "donation"
Are you joking? They just bought HF for $13bill and didn’t break a sweat. They are dominant and spreading.
"The proprietary shovel seller has some excellent tutorials on how to dig gold. Nobody else has such good step by step guides. Therefore them buying a shovel-agnostic tutorials and techniques method (that has a lot of info on using other shovels effectively) is justified."
I'm with Torvalds here, fuck you Nvidia.
NVIDIA is a lot more than a “lab”.
lol “Everything you said is correct but you’re wrong because I’m thinking about something other than what you were talking about”
>No need to quote Linus Torvalds here
Evidently we should, because Linus has been more positive about Nvidia in the last 2 years [0]. I've been using the open driver for years now, for both gaming and CUDA.
[0] https://binarymusings.org/posts/talks/linus-on-ai-linux-in-k...
He definitely sounds more pragmatic than before, which is, in a sense, more positive.
> This is actually one of the benefits brought by AI; it has made Nvidia a good participant in the Linux kernel space. [...] Now, when Linux is so important for AI clouds, Nvidia suddenly cares very much about Linux.
Nvidia releases some of the most open open weights models, Nemotron 3, which have the full training code open, and most but not all of the training datasets.
Nvidia is a big company. They are good about some things and bad about others.
I think they really do like open weights because they make some of the best hardware for training, and the more open weights models there are, the more people are training and fine-tuning them, mostly on Nvidia hardware.
I feel like Nvidia is one of the better choices for buying Huggingface. Not perfect, but definitely far from the worst.
What would be the worst three?
I think this is part of an open-source play. I'm not arguing your other points, I think they're true.
They're trying to mix up the competitive landscape(that doesn't impact their bottom line, and I don't think opensource is eating their lunch), so I don't think this is fake, at least that's my initial take.
Nvidia has hardly open sourced the nvcc compiler, yet no-one here is complaining about why it is closed source.
Modular on the other hand creates the Mojo compiler gets criticised for not open sourcing it immediately and now once they do, no-one cares anymore.
Huggingface was not just a target for open source, but as a force to have open weight models run better on Nvidia against the rest.
> as a force to have open weight models run better on Nvidia against the rest.
this is the crux - if nvidia makes it so that open weights end up running better on nvidia hardware than competitor's, then it's going to prevent hardware innovation and competitiveness in the entire sector.
It's like as tho General Motors buys out oil refinery to make gas for all, but the gas somehow runs smoother in GM cars.
It's a clever move if that's what they're doing. They're restricted in China, and are likely to face stiff competition from Chinese chipmakers in the coming years. Acquiring the largest repository of trainable models and ensuring they run better on Nvidia hardware is probably one of the few moves they have for keeping ahead of the competition. I mean, it would be terrible for the consumer, but it does make me think that NVDA is a decent investment.
I assume it's because nvidia is a hardware company that produces world class hardware, and it's a compiler to target that.
Whereas mojo is a general purpose language, and we're absolutely spoiled for choice on modern languages with open source compilers.
I'm not saying it's fair or right, I still think mojo is neat, but isn't exactly comparing apples to apples.
We can stop with these weak excuses since AMD and Intel have done more for open source than Nvidia has, including their GPU drivers for Linux.
Nvidia on the other hand has not and the best they have done is a bunch of closed-source blobs which they do more closed source releases than the rest.
Mojo is open source and targets all GPU architectures for their compiler regardless of the vendor and nvcc targets their own (and both that and CUDA are closed source).
So this is directly an apples to apples comparison.
> no-one cares anymore.
possibly because it took qualcomm buying them to make that happen.
Just say you don't know.
Mojo was partially open source before Qualcomm bought them, and they were going to do open source it anyway.
Was NVCC or CUDA ever open source since the lifetime of its development?
> they were going to do open source it anyway.
uh huh.
> Was NVCC or CUDA ever open source since the lifetime of its development?
you ever ask Nvidia why? i did.
I for one welcome our new robot overlords.
I think it's a plausible play that keeps the AI boom in session for a year or two longer.
Or at least, $13b to stay at the head of the race (or keep the race running) must be worth it to someone's desk.
There's only $50b in datacenter buildout nationally (Source: Gemini, 2026).
So it is a bit of a puzzling choice for what amounts to a pile of software, in my opinion. but I don't know shit.
Did you just quote an LLM as a citation? That's not how citations work.
"That's not how citations work." (Dude on the interwebz, 2026)
But more seriously, this is my first time seeing that as well, and I'm not sure I like it. Citing an LLM is a little like citing Wikipedia to me, you cite the primary source the LLM is quoting directly, not the secondary source.
A Wikipedia link can be nice when you get several different sources provided.
It’s the formatting and formality that bother me. Say “I looked it up with Gemini,” don’t give me some weak attempt at “proper citation” to legitimize the bare minimum effort you put in.
Why waste keystrokes when less do trick?
Even worse is that Gemini's knowledge cutoff is still Jan '25.
2026 equivalent of “my Mama told me”
> There's only $50b in datacenter buildout nationally (Source: Gemini, 2026).
Your reference lacks authority, veracity, and reproducibility.
It came to me in a dream.
They’re buying a brand, some employees, and some momentum — not any of their software. HF probably does have propriety goodies to make it all run efficiently, but certainly not a billion dollars worth, much less 13!
We are so used to these numbers being thrown around in the AI era that something one needs a reminder that this is 13 billion, not million. Insane exit by HF.
Was looking for this comment, the tech world has gotten completely insane with ”valuations”. I would love to hear why it was 13 and not 5. It would still be completely insane at 5 billion, but someone though they should add another 8…
Yeah, the price is weird as fuck.
Potentially horrible for monopoly reasons, but if other big acquisitions in the AI era teach us anything, developers are about to get a whole lot of free and discounted trial credits.
That’s at least a plus. I will happily burn through as much VC money as they will give me to tinker with my projects.
Yes. Great for devs, for a while. Awful for overall competition and industry's health.
I think the federal antitrust regulators are asleep.
Edit: antitrust regulators' job has just begun -- we'll see how this deal gets adjudicated by the FTC (if at all).
> I think the federal antitrust regulators are asleep.
Have been for a long time. All of the big techs should have been broken up long ago.
You can torrent and seed these models via torrenting. They are not piracy so there's no problem hosting them besides maybe not having enough bandwidth
Interesting ideas, maybe the real open alternative will indeed be torrents.
Except for during Biden, they've been asleep for almost 50 years.
Now that you mention sleeping. I imagine one could do cool sleeping startups that clone something but exist only on paper. The sleeper startup is only launched the moment some ham fisted mega corp aquires the original and all the customers are looking for a way to abandon ship.
Lina Khan @ the FTC was the best thing about the Biden administration, for me.
I think failing to take advantage of the charisma of both Lina Khan and Janet Yellen, and not making them public-facing economics gurus like Alan Greenspan or Larry Summers historically were, was a huge tactical mistake by Biden.
Yellen was objectively the dumbest Treasury Secretary ever. When everyone and their dog realized inflation was getting out of control, she thought it was transitory. She literally gets an F- as an economist.
And some former Roomba employees may have thoughts about Kahn.
Kahn was right about Roomba. Amazon was going to torpedo all the superior competition on the Amazon platform where almost all robot vacuums are sold in the US, plus as the e-commerce monopoly, Amazon was going to instantly have access to an army of cameras inside people’s homes to analyze personal information related to purchase preferences.
IMO allowing Roomba to hit chapter 11 was still a better option for the consumer than handing them to Amazon. They’re still in business as an independent competitor on the market, consolidation was successfully avoided.
Yellen, I don’t have much of an opinion on, but she absolutely wasn’t alone in that opinion (the nuance of that opinion being inflated by this hyperbole), and the treasury is a lot less involved than the federal reserve in doing anything about inflation, anyway.
Roomba was not competitive in the market, no amount of Amazon market manipulation would have changed that. Best case scenario, Amazon would have invested in making it competitive. Worst case, their engineers would have been absorbed into Amazon's warehouse robotics projects.
Is either of those worse than what actually happened: the company is now a zombie brand for a Chinese company?
Of course Amazon would have changed that. You would go on Amazon and search for "robot vacuum" and Amazon would put iRobot at the top of the results. Review manipulation on the platform would be trivial.
Amazon could email/push notification/text customers asking for reviews of iRobot vacuums but then not do the same for competing vacuums, skewing their reviews higher (asking for reviews boosts ratings by gathering opinions from happy customers who usually don't bother writing a review). Competing brands potentially don't even have your contact information to ask for a review.
Go on Amazon right now and search for "usb cable." There's a giant banner at the top that recommends the Amazon Basics brand, three across horizontally, which on my desktop monitor takes up nearly 50% of the screen real estate. Then below it are the Anker cables that you're more likely to be looking for.
They would have almost certainly been manipulating pricing on them as well. For example, they could take the strategy of lowering the price of the vacuums to break even or sell as a loss leader, but use them as a data-gathering robot in your house to help Amazon sell more of everything else. They could make their Alexa smart home platform preferential to iRobot or lock out competing models.
Robot vacuums are a consumer goods category that is heavily skewed toward Amazon.com as the place of purchase compared to other retailers.
I think "zombie brand owned by a Chinese company" is actually preferable, yes. They still operate and sell vacuums competing with the other robot vacuums on the market, and they aren't in service as household data collection bots for a monopoly e-commerce platform.
I don't think the ownership of the company being foreign or domestic is very relevant to the FTC's goal of preserving positive trade conditions. Would we think the iRobot situation was a bad outcome if iRobot was purchased by a foreign company we view more positively like Miele? We only think of it negatively due to anti-Chinese bias. iRobot being Chinese-owned is almost certainly the best possible outcome for preserving the amount of competition in the market.
This is a tangent to my first comment parent to this one, but also more related to the actual article at hand: whether it's Amazon buying Roomba or Nvidia buying Hugging Face, I have a somewhat radical (or is it?) opinion that large companies like Nvidia, Amazon, Apple, Microsoft, Coca-Cola, etc, should not actually be allowed to acquire companies. At all. under any circumstances, even if competition is healthy.
These companies are generally large enough that they do not need the competitive aid of buying an existing company and starting with that sort of structural advantage.
E.g., did Nvidia not have enough money in their bank to start a company to compete with Hugging Face? This is a company that reportedly has ~200-300 employees with investment rounds totaling $400 million. Nvidia made $31.9 billion in net income last quarter.
I look at a company like Xiaomi which just developed its automotive division in-house without resorting to buying car companies. I think our traditional business and finance mindset has an overreliance on acquisitions.
iRobot is now 100% owned by Shenzhen Picea Robotics of China. (https://www.sec.gov/Archives/edgar/data/1159167/000115916726...)
Yes, that's a very good company to be the owner of iRobot, because they are just a robot factory and not a near-monopoly e-commerce retailer.
They are essentially on equal footing with other robotic vacuum manufacturers. iRobot didn't go out of business or get absorbed into a larger company lowering competition in the marketplace.
It's also not the FTC's job to ensure that companies, especially ones with zero/trivial national security or domestic labor force value, remain under domestic ownership. Amazon itself is not really a "domestic" company, it's publicly traded. Anyone from any non-sanctioned country can buy shares of Amazon.
How would you define a domestic company?
The company's name was iRobot, not Roomba.
I spent a little time working there around 2017. Whatever you're implying my former colleagues' thoughts on the subject are (or mine, for that matter), you're probably wrong.
I miss the Biden administration
Large numbers of people outside of the United States probably do.
I miss him from inside the United States too.
China and Russia are loving things right now.
I miss every administration back to Roosevelt.
That's a sad thing to have to say, but yeah.
True, but better nvidia than any other big tech contender. Nvidia has a very strong business incentive to make huggingface thrive, whereas pretty much everyone else has the opposite incentive.
Especially when all major labs are building their custom inference chips.
The Government plans to regulate open source ai. Nvidea is dependent on the Government to back loans, provide property, and make laws in their favor. Now they have ownership of what the Government wants to control. A perfect wedge they can use to get more from the Government.
We're entering consolidation
> That’s at least a plus. I will happily burn through as much VC money as they will give me to tinker with my projects.
The era of VC-funded home-delivery recipe boxes is still my favourite.
Yeah this is the "early netflix" era. Take advantage while you can and plan for the future.
Counter point is China can afford to keep pumping out free models for long enough to crash the AI industry, As well as spin up memory production.
Its likely in the near future we will be able to buy 128gb mac minis and run local AI for free.
When did apple stop charging more for their memory upgrades then gold by the ounce costs?
Why would China want to crash the AI industry? Their goal is not to damage the United States, necessarily. Their goal is to have the best domestic AI on the planet. They intend to do that through the way that they dominate every other industry: good enough quality at a much greater scale.
If we survive, that is.
That means we will torrent these ai models. They are not pirated so there's no problem doing so
Nvidia is a public company. What VC money?
I think this is good.. They have incentives to keep things free to keep people using their GPUs. I think that's one of the least enshittifying outcomes possible
Well congrats to Clem and the team. I remember when huggingface was doing things like coreference resolution models on spacy.
I hope nvidia does right by the community.
Edit to add: $13B should cover the S3 egress fees for a couple months :D
Totally, and as long as they don't turn on the AWS NAT Gateway, they'll be just fine.
They need to switch to egrees free cloudflare R2!
The $13b doesn't go to hugging face's bank acct to pay for anything like S3 egress fees... It go to hugging face's owners' Bank accounts for them to do anything else. New cohorts of billionaires and centi millionaires getting minted.
Obviously, NVIDIA is trying to own the AI development chain.
Owning HF -- the discovery and distribution channel -- is one thing, but I think the biggest threat vector is the privileged access to HF platform data, that includes HW survey info and model download pattern. This can be a borderline anti-trust case.
If the acquisitions of late tell us anything at all, it's that Nvidia wants to be a vertical AI company, not a hardware or gaming company.
This whole thing made the AI attack on HF even more suspicious...
The more interesting is HF turned down a $500M Nvidia investment late last year at a $7B valuation, after passing on a $235M round in 2023 at $4.5B — going from "we don't want a dominant investor" to a $13B full acquisition in under a year is quite the reversal.
Well, if they didn't want a dominant investor, they didn't get one, they got acquired.
Sounds like it was just too much money for the stakeholders to refuse
Yea, my bet is that there is quite a long list of money bags sitting behind the leadership team that will weigh in if the offer is juicy enough.
I don't think they passed on the 4.5 billion. They still raised it in 2023 with investors like Alphabet etc. They turned Nvidia *earlier this year* for the $7 billion valuation deal. It makes sense if they were angling for an acquisition.
This is the article you are referring to is this but your facts are wrong: https://techcrunch.com/2026/08/24/hugging-face-reportedly-in...
Seems the opposite of a reversal to me.
Reversal would be "valued at 7B then valued at nothing", this is more like "nah, we want more" then "nah, we want more" then "yes, that's what we want :)".
I have never seen an acquisition be good for the users. I struggle to believe this one will be different
I'm sure it'll be fine. As long as you have the latest 90-class card you'll probably continue have nearly full functionality from the HF libraries. (Upgrade to enterprise gear for full functionality)
The fit is theoretically good (nvidia is model agnostic) but I can see then pushing the ecosystem increasingly towards cuda and some custom nvidia software packages for inference etc etc
Apple buying Workflow (became Shortcuts). Its users have a better, faster tool and got it soon after acquisition. And it has many more users now.
Was Workflow available on every platform running on every hardware provider?
I’m not sure your point, sorry.
I think that it was already iOS only, so making it free was just flipping a flag etc. I'm not sure how that helps his point tho.
Acquisition of GitHub let to increased downtimes, but also to unlimited private repos for the free tier.
Overall GitHub still got significantly worse since the acquisition. Downtime was essentially zero before.
Usage was 5% of what it is now before.
The unlimited private repos for free tier would have happened anyways. You have to remember that this was around the time people were jumping ship from GitHub exactly because they didn't offer private repos for free. It was something many wanted, and they were actively seeking out other places that would give it to them.
... Which gitlab already had before github. And still does.
Truer words were never spoken. This is painfully true.
Yeah I'm afraid they will only keep cuda support for their software. Good thing it's all Open source
Remember just 6 months ago that "Ggml.ai joins Hugging Face to ensure the long-term progress of Local AI" (https://news.ycombinator.com/item?id=47088037)
(Ggml.ai is llama.cpp.)
Curious if the “I consider HuggingFace more "Open AI" than OpenAI” sentiment in that top comment will still apply with NVIDIA as the boss now...
Nvidia is promoting "unmetered intelligence". Jensen wants everyone to have a spark dgx with an nvidia chip in their home.
Maybe but they make their real money in data centers. Certainly, they want to have fingers on every trend. Owning something the Government wants to control gives you more access to trade repression of open models for more Government favors and money.
NVidia has reasonable incentives to keep things open indefinitely, it wants people to use it's GPUs
Issue is that llama.cpp is the best way to run models on hardware that isn't nvidias.
Except when they have less than 16 gb of ram?
It'd be really nice if I could use my egpu 4090 on my macbook pro..
What is the business model of hugging face? My understanding was they were basically just a file hosting platform.
Didn’t we just witness it? Become a critical ecosystem player then get acquired based on the value of that position.
I’m asking more to work out what is the basis of this valuation. Why would Nvidia spend $13B for what seems to be a services that gives things away for free.
Talking about profit is so passé in the new economic paradigm. The rules have changed— it’s about how much a company is worth. Get with the times.
–Some guys in every bubble I’ve witnessed.
I think Russ Hanneman calls it a "pure play" type of company. He put radio on the internet, so he would know.
I dunno. You can argue over whether they're overpaying, but it's not like Huggingface is Clinkle. They hit $150 million in ARR this year, they have tons of runway, and according to reports, have just started to even burn the money they raised a few years ago.
I get that it's fun to be glib about the stupidity of tech elites and investors in general, but Huggingface have been pretty open about their financials and are, in my opinion as a practitioner in the field, one of the most responsible orgs in our space. They've been a pillar of open source ML for years now and have made a very positive impact on our ecosystem.
Nvidia is getting a real business generating revenue, and the center of the universe for open models. Both seem like pretty valuable attributes, from Nvidia's perspective.
150 million a year? Why, at that rate NVidia will make their money back in just 86 years.
Solid point. I'm sure Nvidia went into this deal expecting completely flat growth and no other benefits to their core business. Sorta like how Meta never increased Instagram's revenue from $0 and is still waiting for it to pay off that billion dollar acquisition price.
Or like GitHub, which was generating something like 200 million in ARR and had never hit profitability when Microsoft bought it for $7.5 billion back in 2018. I'm sure it has come as nothing but a happy surprise to Microsoft that GitHub generated $1 billion in 2023. They had initially penciled it in for 38 years til ROI.
This gets mentioned a lot, but the most valuable companies, by orders of magnitude, have the most profit, by orders of magnitude.
Risk management.
Nvidia has a market cap of $5T USD today, and a decent chunk of that is due to LLM speculation.
Does Nvidia want their stock price to be at risk of being tanked by a download service being in the news? No, they want to make sure the party keeps going and is under their direct supervision, and part of that is making sure Hugging Face isn't bought by a competitor or runs out of money.
> Why would Nvidia spend $13B for what seems to be a services that gives things away for free.
To stop that! Literally. To stop those services being free.
Why would Nvidia destroy the first place we all look for models to load on our Nvidia hardware?
for 6 months?
This is like saying GitHub or Google just gives away everything for free
Not really. GitHub has a strong platform of paying customers and a solid product that’s hard to replicate.
Google is obvious.
Hugging Face is a file download mirror with a couple of side features dangling off.
They hit 150 million in annual recurring revenue this year.
Pretty nice dangling side features apparently.
That still leaves the company valued at 86 years of revenue
Yeah, great point. Nvidia could potentially be overpaying. Not sure how that equates to Huggingface being "a file download mirror with a couple of side features dangling off".
Absolutely none of this makes sense and the thing that amazes me is how long it has continued. Future historians will just laugh at how stupid and obvious the crash was.
Just wait until you hear that Facebook bought Instagram for $1 billion in 2012, when it was generating no revenue.
While it's probably easier to say this in retrospect, they were eliminating a direct competitor to their core business. Something so advantageous it should have been blocked by regulators.
I can't see this as being as good of a purchase, especially when it's 13x the price of what was seen as an absurdly large amount back then.
You are very good at bringing up irrelevant numbers and missing the forest for the trees
Whether you realize it or not Hugging Face's monetization is just as obvious as GitHub's
They make it easy to run models. The models have to run on hardware somewhere.
https://huggingface.co/pricing
Possibly, but they were profitable already.
They have various paid services that relate to AI—paid inference hosting; paid accounts aimed at AI development with GPU rentals [credits plus paid overages, I believe], more private storage and public storage than free accounts; and many of the community a and some other benefits on individual/team/enterprise tiers; additional paid storage above the base quotas for the paid account tiers; on demand rentals of HF managed containers on GCP and AWS, and some other things.
Isn't this pretty much the model for any tech company? Rush to get MVP released, stoke the hype, cash in on FOMO.
It hugs your face
If GLM-5.3 was truly trained and ran on Chinese chips, we'll have to ask what the business model of Nvidia itself is soon :)
I guess we may have to start paying to download models.
Or perhaps they will start throttling downloads for free users.
I don't know what they business case is, it might be to shut them down: I suspect good free models on local hardware is a threat to Nvidia's investments in OpenAI/Anthropic.
I see the opposite — NVIDIA is aiming to commoditize the model layer and push open models that are tuned to run on NVIDIA hardware.
NVidia has been expending energy helping improve local models and inference platforms for them targeting NVidia GPUs; good free models that users can run locally rewards Nvidia’s investment in product lines for local inference (DGX, RTX PCs, etc), as well as—given their continued dominance in the space—the premium over competitors of their consumer and workstation GPUs.
Maybe everyone will migrate to Hugging Bay for downloading models via torrent?
File hosting isn’t that hard to replicate though. Even if they start charging people will just set up torrents for the files or lists of mirrors.
Unless I’m missing something, this feels like Nvidia having more money than they know what to do with.
Maybe they're trying to see how big the market actually is before considering shutting it down, maybe or getting lawyers and politicians to try and outlaw or restrict open models if they see a big enough opportunity.
it's literally the GitHub for AI
Github was acquired for $7B. It's hard for me to see how HF is worth almost double Github.
GitHub was acquired for 10B in 2026 dollars. And software on HF requires quite a bit more hardware expenditure to run.
ondemand cloud? b2b?
The open weights AI ecosystem is way too concentrated with HF. That said, it was clear they were going to either get bought or aggressively monetized at some point, Nvidia seems like it could be one of the more aligned acquirers.
I do worry about any sort of crowding out or downplaying non Nvidia-relevant quants, and about changing rules to crack down on models or datasets that for one reason or another “don’t align with their corporate values” - uncensored etc. Someone mentioned Microsoft and GitHub, they appear to me anyway to have been very hands off, I hope it’s the same model.
There are always torrents, which would seem to be a fine way to distribute models.
We also have modelscope.cn, I'm sure that one won't get acquired
It's already part of Alibaba, which has its own incentives. So does kaggle with Google. Civitai is neutral but it is mostly nsfw even though they don't show it outright. Torrenting is the Best way
HF is the default platform to find models. Not just ones published by big labs but also a lot of the distilled or fine-tuned versions, etc.
If Nvidia buying HF makes it tough for all the diverse models on HF, then what are some alternatives?
It seems models are the best things to be available on a Torrent platform? Of course HF is much more than just the files but perhaps the metadata can be separate and hosted on multiple community platforms.
If HF didn’t exist, model releases would likely have been on torrents like Linux iso’s back in the day.
I'd honestly prefer that. I'm usually on 5g if not 4g and pulling models is an absolute nightmare when I basically have to use HF and even with a token getting throttled. Although I have a pretty unique use case. Pulling wads of 5gb tensor files over cellular is painful.
I am on the same boat, I live in a small Himalayan village. I have 2x5G based devices and one Wireless bridge (Ubiquiti LiteBeam M5) for a local broadband. Generally I get 50 Mbps, sometimes up to 100 Mbps
> what are some alternatives
good old torrents
There's over 3 million models on HuggingFace. So many quants. I'd imagine less than 0.5% would actually be seeded.
This is the world we live in now. 3-4 megacorps owning everything in every major product category - eyewear, cosmetic, consumer goods, media, tech; you name it.
We need small companies worldwide, developing the things huggingface does. The us china and Europe dominate while the rest of the world sits by idly. Torrent and seed ai models. It's not piracy
If you can't beat them, buy them. Hallmark of late-stage capitalism.
I really hope they retain and pay Georgi Gerganov. We are toast without ggml.
I am just curious, and I genuinely want to know, how is a 150 million a year in revenue company worth 13 billion dollars?
If they are mostly paid in NVDA stock and NVDA is also valued at 40x revenues or whatever insane multiple then it makes sense.
> Here have some of my monopoly money I can print and come join us at Nvidia!
Somewhere, there’s a slide deck that shows future revenue going hockey stick.
Curvature and convexity matters even in "hockey-stick" models =3
https://www.youtube.com/watch?v=NufJ7g63KSY
Same reason why GitHub was acquired for $7.5B at $250 million ARR which is 30x revenues:
VCs could not see any other reason to raise more money and Huggingface was not growing as fast as they thought to justify the valuation or the next fundraise.
So they might as well get Nvidia to save them from the VCs pressurizing them.
That doesn't make much sense to me. Companies don't just buy companies for absurd multiples out of the niceness of their heart.
In the case of GitHub, it was likely for data reasons + wanting to own where developers do work (VScode + Github).
In the case of HuggingFace, honestly not sure as I'm not familiar enough with their business. But I can assure you that Nvidia didn't buy them for 13 billion cause HuggingFace were desperate. When you're desperate, you sell for less not more.
I remember when people realized gaming cards could be used for general compute (GPGPU) on certain types of parallel numeric work. It was a very cool thing for many types of hackers and science and engineering students and NVIDIA went far out of their way to ensure we could never buy it.
The optimal market strategy there (as in a lot of places) wasn't "sell as much as you can". There's often a superior strategy, when (as with HPC) you have minority industry customers who are very rich and have low price sensitivity. It's to raise the price to what those special customers are willing to pay, and to drop everyone else.
What NVIDIA did was to rip out FP64 capability, systematically, from all of their consumer cards. They firewalled off "useful for GPGPU" as a differentiating feature, segmented the market, and astronomically raised the price of what (if you were looking soley at cost-to-manufacture) could have been easily affordable to any ramen student.
(It's a more obscure version of the Intel-made-ECC-memory-disappear story).
See, e.g.
https://news.ycombinator.com/item?id=47068890 ("15 years of FP64 segmentation, and why the Blackwell Ultra breaks the pattern (nicolasdickenmann.com)")
The same is true for the simultaneous NVENC video encoding session limit which is artificially locked for consumer NVIDIA GPUs (see https://github.com/keylase/nvidia-patch)
Yes, this is ironic - it was the tinkerers using CUDA on cheap graphics cards that made Nvidia stuff useful in constrained academic environments.
True about Intel and ECC, but AMD now does similar things, even with their consumer CPUs and chipsets.
These three companies now make very sure that consumer products can never canibalize those juicy data center profits - so they make sure to limit what the consumer segment can do.
The companies have aligned incentives so I don't think this is a terrible acquisition for customers.
Vertically integrated monopolies naturally have aligned in incentives for themselves, which allows more capture which turns out bad for customers. Whether it is a monopoly is an open question.
This isn't vertical integration. This is a commoditize your complements play and that does benefit customers. Nvidia benefits from supporting the open model ecosystem, if models are a commodity the gpus become what's scarce.
of course it is
works fine for now
whats nvidia gonna do except make it worse?
usually buyouts go something like this:
1 buy company
2 fire various people
3 enshittify
hugging face even said they didnt want to accept a 500 million dollar investment from nvidia, because they didnt want nvidia to run the ship.
instead they sell it... guess who'll run the ship?
https://techcrunch.com/2026/08/24/hugging-face-reportedly-in...
Is there any cash involved or is this all stock?
stripe buys openrouter. nvidia buys hugging face. everyone on the hunt for the app that will function as hub/broker/search/marketplace for models.
I hope they're good stewards. This reminds me of the Microsoft GitHub acquisition in terms of the importance to the broader community.
As someone who've played video games and obsessed over graphics cards long before this all started, don't get your hopes up.
as gamers, we got burnt by nvidia way too many times. I'm still holding grudge over shutting down 3dfx.
Nvidia is indeed a terrible company when it comes to open source, etc. however, as someone who has been gaming on desktop hardware since the 6xxx card days, Nvidia hardware has always worked. I switched to Linux full time since the 2XX days (maybe it was the 280 gtx?) and never had a lot of the problems everyone else got on Linux (using only the closed source drivers).
NVIDIA introduced programs like The Way It's Meant to be Played, optimizing major game titles specifically for NVIDIA architecture, which made competitions like AMD cards underperform in critical releases.
I'm old enough to remember.
I fail to see how this is a problem?
AMD could’ve done the same.
Amd and Intel have the same programs.
But they don't/didn't have the same level of market share as Nvidia, particularly in the PC gaming market. Nvidia had 65%-85% market share, and still worked with game studios to make sure their competitions failed. That's the opposite of good stewardship that's all I'm saying.
You are being disingenuous.
They worked with studios to optimize games for their hardware. I don’t think they went around sabotaging other companies hardware. They did what was good for them.
Why did competition not do the same?
Likewise, their hardware has been rock solid and very performant for me. I bought most of my current setup second hand and they've been going strong for years now.
3dfx was going through bankruptcy
I think they will be, Nvidia has released open source model that also included the data it was trained on. I think they are the only ones that have done that.
Handing out digging spots if you are the shovel manufacturer sounds reasonable.
As long as the spots continue to be compatible with shovels from other shovel manufacturers
They were already pretty vocal about it, but this makes Nvidia pretty much the de facto face of open weight models.
As battle lines get drawn over duopoly vs. open weight it’ll be interesting to see what Nvidia does. They definitely want a piece of more of the stack especially as Huawei chips become more and more of an alternative to cuda.
Alibaba already has modelscope and their own chips, and qwen models. Nvidia is doing the same with chips, huggingface and nemotron models
NVIDIA had more open-source models on huggingface than any other company. A lot better them than any other frontier lab, tbh.
Monumental implications on OS models. Proper investment in US OS incoming? HF has the best possible distribution to that audience
Like some other folks, I wonder whether this will be a good thing, or a bad thing.
I feel like if Nvidia ends up turning into a bad actor, in terms of restricting/censoring models... another HF will spring up.
Fix your title it's not finalized yet!
According to The Information, it seem it's finalized? https://www.theinformation.com/articles/nvidia-agrees-buy-op...
Then use that as the link.
Everything I can find online is referring to this one source. That doesn't tell me if it's happening or not.
Oh no. I use huggingface all the time. Sad news.
Not good news and in the long run this circular move will fail for them. What are the alternatives?
The lowest hanging fruit that hf could offer is community monitored commands for running local models.
Right now it is a pain to find the correct incantation.
What does Linus’s hatred of NVidia drivers imply for this situation?
Looks like owning the hardware wasn't enough and they want to own the ecosystem as well
Great for Nvidia. Bad for the rest of us. Oh well, so much for hugging face.
Huggingface sold out. Sad day. All the good companies lose their 'dont do evil' clause as soon as they make bank
You've sold out when you became VC Funded... It's just a matter of how long the free punch lasts.
It was either they (literally) sell out or enshittify. Choose your poison. Hosting and running LLMs is an expensive business to be in. There wasn't ever going to be a good ending to this story. The original investors need to be made whole.
What, are you expecting them to just work for good vibes?
To be fair, this is the best thing that can happen for the founders. I don't blame them.
Brilliant replacement of http://build.nvidia.com to increase developer reach and deployment of nvidia open models!
No, what? Did they just do this to make Chinese models harder to acquire outside of China?
They will just download them from modelscope
What’s with all the big acquisitions recently? Is it advantageous for some reason to announce all of this stuff now?
The VC capital for AI is probably drying up. Now the question is: who will buy OpenAI and Anthropic?
The public
Not if they can't indefinitely contain ram and SSD procurement. And they can't and the Market already showed with specex that they are unwilling to float these fly by night outrageous 'investments' onto institutional holders.
Institutional holders mass revolted at spacex getting into the basket.
AI costs more than having people do the work. Q2 CFO reaction proved that.
Congrats to the team. I have a strong feeling this will go down in history as one of the last great exists before the bubble bursts.
12-24 months from this acquisition will likely look like a crazy burn of capital and cash.
That's a great news for a few rich wallets and bad for all of us.
I'm puzzled by the reactions. This is not a good thing. Yes, they're releasing more open source code and models, but they do it BECAUSE OF pressure from the community and other open source projects.
But Nvidia is a terrible open source and consumer company. They gatekeep a lot and oftentimes it's only open source in name. Outside contributions are often slow-walked or rejected if they don't align with business incentives , and leadership is retained 100% in a couple of people from a certain country.
Linus Torvalds: 'Nvidia has been the single worst company we've ever dealt with'
His other comments on Nvidia often contain expletives.
Linus isn't wrong but at the same time NV has shipped more compute than AMD and Intel combined to consumers. The kind of performance I get out of a modest outlay never ceases to amaze me.
Sales volume has no correlation with how good a company is for an ecosystem. See Microsoft, Adobe, Apple, Sony, and so on.
NV is a very different beast from all of those. I have yet to regret buying any of their products and I wouldn't touch your list with a stolen 10' pole.
In fairness, most of those comments were made before Nvidia capitulated in the EGLStream war and released open kernel modules.
Isn't it super easy for the community to just replace HuggingFace? Isn't it pretty much just a repo/index for open-weight models? Why is it even worth anything?
> Isn't it pretty much just a repo/index for open-weight models?
No, its not just a repo/index (they also have training, inference hosting, and they develop/maintain a bunch of core AI infrastructure software), and even if it was just a repo/index, replacing a bug centralized repo/index that used by an large community isn’t trivial.
Not just a repo/index but they also offer reasonably-priced compute, with a free tier not to sniff at, on which to run the models.
Network effects. It's like a social network for LLM nerds as well as hosting and rating models, etc.
It also means that finding models that aren't "approved" will be dealt by other websites.
good or bad, it comes down to a financial decision. Nvidia was willing to offer the highest bidding price, and HF was willing to sell.
I wonder if OpenRouter being acquired was a catalyst for this deal.
Monumental implications on OS
Boy, there sure is a lot of money flying around...
First they came for all the dev tooling - uv, Cursor, etc. Now the routers and providers - Open Router, Hugging Face...
Who or what is next? And what is the endgame I wonder??
welcome nvidia :)
So is the next step to just basically create a cloud business.
they basically own chips - data centers - discovery etc?
Its becoming impossible to track all the acquisitions.
I'm more worry than positive tbh.
Nvidia has already a good free offer with Nim and a lot of compute power available. Plus, they obviously have interest in spreading open models culture.
We keep talking about EU having to be more present in the AI race, but if the lifecycle of European AI companies is to be bought by US ones, what’s the point?
Hold on, these say no deal has been reached yet.
Best case scenario
HuggingFace offers abliterated/uncensored models, which would seem to run contrary to the "AI is too powerful for regular people" narrative that the big players are pushing.
My primary concern is that Nvidia will bow to the pressure and restrict abliterated/uncensored models on HF.
Pretty sure Nvidias dream is that OpenAI and Anthropic get into a hacking war using Nvidia cards.
The problem for them is that the leading provider of training and inference is actually AWS...
That is a very good point. I will set aside some storage and download those models that I think are most at risk from a corporate clean-up action.
I don't think nvda has ever pushed the idea that ai is too powerful for regular people.
That's the anthropics and openais of the world.
Wow, nice strategic move and price seems a bit low tbh. Good deal.
I’ve asked ChatGPT etc a few times how Hugging Face makes money. Never quite got to understanding . Good for them.
You could just go to their website and look at the pricing pages? Why are you asking an LLM?
Because I never ever see anyone talking about using it, except for occasional reference to weights.
Gemini did a good job of explaining it. fwiw, it may have helped that i accused Gemini of leading me to a "scam site" like "mega dot com"
Big Wolf buying the sheep farm. Nice
nvidia antitrust incoming
I wanted to pray for AMD to attack somehow but then I remembered their owner CEO's are cousins. Should we pray for Chinese GPUs?
Oh jeez
I really wonder why Nvidia would do this.
Owning the roots, the trunk, the branches and the tip of the leafs.
whoa
This is huge!
s-1.site
Just like that they lost their ethos. AI oligarchy in the US is now cemented.
I really wish I'd done better bookmarking back when Nvidia was talking about acquiring ARM. There were all kinds of things coming out saying, if we do this, even though we will be taking over ARM, it will transform us from the inside out. We would become a new different company, that cares about something beyond our own self interest, our own chips.
This feels like a similar leap of faith. One that is hard to believe in. Thankfully, I think Nvidia can keep the lights on here & keep this going. I don't think they have to do much, per se. But it felt implausible then to image an Nvidia that gave a shit about anyone else, an Nvidia that actually gave a flying fuck about drivers or upstream Linux or ecosystems that weren't entirely within their own control.
Similarly the upper quartile of succes here feels mostly like benevolent neglect. I think we can hope for Nvidia to just not mess up a good thing, for them to understand that this open model open ai universe hinges upon Hugging Face, and for them to pretty please keep caring about the existential risk of the hyper-ai'ers all building their own properietary models on proprietary hardware and leaving Nvidia behind some day, and HF being the hedge against being left behind.
um um um um do we want nvidia to be on the same network as hugging face given the recent incidents?
the title of this post is misleading… “in talks” != not “agrees”
It’s not misleading according to the The Information piece (second link)
The story is in an intermediate state right now, but as semiquaver points out, The Information is reporting it as fact. Since their reporting tends to be as reliable as it is hardwalled, we went with their claim in the title.
Anyone old enough to remember anti trust laws or even just laws?
Please don't post unsubstantive comments to Hacker News.
No doubt you have the nucleus of a substantive comment here, but that's not enough. If you only post the shallowest top stratum of what you're thinking, other people do the same, and then we get "Laws are for poor people" and endless descending repetition. The whole point of this site is to try for something other than that.
https://news.ycombinator.com/newsguidelines.html
p.s. Also, please don't be snarky on HN. That's also in the guidelines.
You are right but its the initial shock on bad news, hopefully this will spawn alternative competing services.
I thin HF holds huge power in how they’ve consolidated all the open models and data, but monopoly feels like a stretch here. They are in reality a fairly small startup with lots of alternatives (all of which are much worse at this point granted) and are mostly a loss leading part of the open source ecosystem. I think it would be very hard to make a case against this on monopoly grounds, though I wouldn’t be opposed to someone trying if they thought it was possible.
Laws are for poor people
The AI judge doesn’t have that feature yet
tldr: the marketplace for opensource models became wide open